CN110348347A

CN110348347A - A kind of information processing method and device, storage medium

Info

Publication number: CN110348347A
Application number: CN201910580576.2A
Authority: CN
Inventors: 阎旭阳; 干刚; 张恩龙; 李冠亮; 曾咿人
Original assignee: Shenzhen Sensetime Technology Co Ltd
Current assignee: Shenzhen Sensetime Technology Co Ltd
Priority date: 2019-06-28
Filing date: 2019-06-28
Publication date: 2019-10-18
Also published as: SG11202108349UA; JP2022518469A; TWI743835B; TW202101444A; US20210357624A1; WO2020259099A1

Abstract

This application discloses a kind of information processing method, device and storage mediums, wherein the method includes: to obtain the first input information；Wherein, the first input information includes at least the image containing target object；Capturing the image collecting device to the target object N second candid photograph images, object time point before and after object time point based on the first input acquisition of information is time point of the described image acquisition device candid photograph to the target object；The same pedestrian of the target object is determined from the candid photograph image；The same pedestrian is analyzed based on polymerization file data, obtains same pedestrian's recognition result, the same person is corresponding with unique archives in the polymerization file data.

Description

A kind of information processing method and device, storage medium

Technical field

This application involves the information processing technologies in computer field, and in particular to a kind of information processing method, device and Storage medium.

Background technique

Public security department is when daily carry out some cracking of cases, it is likely that the face picture without target suspect occurs And other are conducive to the case where relevant information solved a case, the personnel's analysis file work carried out at this time for this person is difficult to out Exhibition.But sometimes, offender can with clique commit a crime form implement criminal activity, i.e., sometimes target suspect have it is suspicious Same administrative staff.When some criminal gang is obstructed or is required to look up to the clue of suspect, by searching for the same administrative staff of suspect It can be to solve a case to provide effective clue.Therefore, the same administrative staff for how determining suspect, is a technical problem to be solved urgently.

Summary of the invention

In view of this, the application provides a kind of information processing method and device, storage medium, target object can be quickly identified Same pedestrian.

In a first aspect, the embodiment of the present application provides a kind of information processing method, which comprises

Obtain the first input information；Wherein, the first input information includes at least the image containing target object；

The image collecting device to the target object is captured in object time point based on the first input acquisition of information Front and back N seconds candid photograph image, the object time point are time point of the described image acquisition device candid photograph to the target object；

The same pedestrian of the target object is determined from the candid photograph image；

The same pedestrian is analyzed based on polymerization file data, obtains same pedestrian's recognition result, the polymerization archives The same person is corresponding with unique archives in data.

It is optionally, described that the same pedestrian is analyzed based on polymerization file data in above scheme, obtain same pedestrian Recognition result, comprising:

The relevant information of all same pedestrians is determined based on polymerization file data；

Wherein, with the relevant information of pedestrian, comprising:

For failing the same pedestrian of real name, relevant information is included at least in system in first database about the same pedestrian Each candid photograph image；

For the same pedestrian of real name, relevant information is including at least the image information in the second database in system, text Information.

It is optionally, described that the same pedestrian is analyzed based on polymerization file data in above scheme, obtain same pedestrian Recognition result, further includes:

Determine the number of going together of all same pedestrians Yu the target object；

All same pedestrians are ranked up based on colleague's number to obtain colleague human sequence.

The first same pedestrian is determined from the colleague human sequence；

Determine the target object with described first with pedestrian all records of going together；

Wherein, it is described colleague record includes at least: the target object with described first with pedestrian candid photograph image, capture Time, image collecting device identification information.

K same pedestrians are determined based on the colleague human sequence；Wherein, K is positive integer；

Determine the target object and the K all records of going together with pedestrian；

Wherein, colleague's record includes at least: the target object and a candid photograph image with pedestrian of the K, candid photograph Time, image collecting device identification information.

In above scheme, optionally, the method also includes:

Obtain the specified video stream by specifying image acquisition device；

The target object and the K same pedestrians under the designated stream are searched from all colleague's records Colleague record.

Based on the target object and the K all records of going together with pedestrian, count in each image collecting device Under the K with pedestrian candid photograph number.

In above scheme, optionally, before the first input of acquisition information, the method also includes:

Clustering processing is carried out to the image data in first database and obtains clustering processing result；Wherein, first number The portrait figure captured according to library based on image collecting device at；

Polymerization is carried out to the image data in the second database to handle to obtain polymerization processing result；Wherein, second number It is formed according to library based on the image information of real name；

The clustering processing result is associated analysis with the processing result that polymerize, obtains polymerization file data.

In above scheme, optionally, the image data in first database carries out clustering processing, comprising:

Go out face image data from the image data extraction in the first database；

By the face image data divide into several classes, if each class in the Ganlei has a class center, and institute Stating class center includes class central feature value.

In above scheme, optionally, the image data in the second database carries out polymerization and handles to obtain at polymerization Manage result, comprising:

The identical image data of identification card number is polymerized to an image library；

The incidence relation for establishing described image library text information corresponding with the identification card number obtains polymerization processing knot Fruit, each identification card number corresponds to unique file data in the polymerization processing result.

It is optionally, described to be associated the clustering processing result point with the processing result that polymerize in above scheme Analysis, comprising:

Class central feature value each in first database and each reference class central feature value of the second database are carried out complete Amount compares；

Similarity highest is determined based on full dose comparing result and similarity is greater than in the object reference class of preset threshold Heart characteristic value；

The corresponding target portrait of object reference class central feature value and the mesh are searched from second database Mark the corresponding identity information of portrait；

It establishes the corresponding identity information of the target portrait and is worth corresponding image with class central feature in first database Incidence relation.

In above scheme, optionally, the method also includes:

When increasing image data newly to the first database, clustering processing is carried out to newly-increased image data, will be increased newly Whether face image data divide into several classes in image data, inquiring from the first database has and newly-increased image data phase Same class is then merged into the existing archives of respective class if there is identical class；If being based on new class without identical class New archives are established, are added in the first database.

In above scheme, optionally, the method also includes:

Whether when increasing image data newly to second database, inquiring from second database has and newly-increased figure As the identical identification card number of data is then merged into the corresponding existing archives of the identification card number if there is identical identification card number In；If the identification card number based on newly-increased image data establishes new archives without identical identification card number, add to described In second database.

Second aspect, the embodiment of the present application provide a kind of information processing unit, and described device includes:

First obtains module, for obtaining the first input information；Wherein, the first input information, which includes at least, contains mesh Mark the image of object；

Second obtains module, for capturing the Image Acquisition to the target object based on the first input acquisition of information Device N seconds candid photograph images before and after object time point, the object time point are that described image acquisition device is captured to described The time point of target object；

Determining module, for determining the same pedestrian of the target object from the candid photograph image；

Processing module analyzes the same pedestrian for being based on polymerization file data, obtains same pedestrian's recognition result, The same person is corresponding with unique archives in the polymerization file data.

In above scheme, optionally, the processing module is also used to:

Wherein, with the relevant information of pedestrian, comprising:

In above scheme, optionally, the processing module is also used to:

The first same pedestrian is determined from the colleague human sequence；

In above scheme, optionally, the processing module is also used to:

In above scheme, optionally, the method also includes:

Obtain the specified video stream by specifying image acquisition device；

In above scheme, optionally, described device further include:

Archives establish module, are used for:

In above scheme, optionally, the archives establish module, are also used to:

Go out face image data from the image data extraction in the first database；

In above scheme, optionally, the archives establish module, are also used to:

The third aspect, the embodiment of the present application provide a kind of information processing unit, and described device includes: memory, processing On a memory and the computer program that can run on a processor, when processor execution described program, is realized for device and storage The step of information processing method described in the embodiment of the present application.

Fourth aspect, the embodiment of the present application provide a kind of storage medium, and the storage medium is stored with computer program, When the computer program is executed by processor, so that the processor executes information processing method described in the embodiment of the present application The step of.

Technical solution provided by the embodiments of the present application obtains the first input information；Wherein, described first information is inputted at least Including the image containing target object；The image collector to the target object is captured based on the first input acquisition of information N seconds before and after object time point candid photograph images are set, the object time point is that described image acquisition device is captured to the mesh Mark the time point of object；The same pedestrian of the target object is determined from the candid photograph image；Based on polymerization file data pair The same pedestrian analyzes, and obtains same pedestrian's recognition result, the same person is corresponding with uniquely in the polymerization file data Archives；In this way, can quickly identify the same pedestrian of target object by analyzing automatically multiple candid photograph images, and due to polymerization File data is established based on one grade of a people, helps quickly to determine the relevant information with pedestrian.

Detailed description of the invention

Fig. 1 is a kind of implementation process schematic diagram of information processing method provided by the embodiments of the present application；

Fig. 2 is colleague's number query result schematic diagram provided by the embodiments of the present application；

Fig. 3 is for target object provided by the embodiments of the present application and individually with the record queries result schematic diagram of going together of pedestrian；

Fig. 4 is same pedestrian's appearance point position provided by the embodiments of the present application query result schematic diagram；

Fig. 5 is the analysis result schematic diagram provided by the embodiments of the present application about single video source；

Fig. 6 is the algorithm principle schematic diagram of face cluster provided by the embodiments of the present application；

Fig. 7 is the implementation process schematic diagram of face cluster provided by the embodiments of the present application；

Fig. 8 is the result schematic diagram of face cluster provided by the embodiments of the present application；

Fig. 9 is archives Establishing process schematic diagram provided by the embodiments of the present application；

Figure 10 is a kind of composed structure schematic diagram of information processing unit provided by the embodiments of the present application.

Specific embodiment

The technical solution of the application is further elaborated in the following with reference to the drawings and specific embodiments.

The embodiment of the present application provides a kind of information processing method, as shown in Figure 1, the method specifically includes that

Step 101 obtains the first input information；Wherein, the first input information includes at least the image of target object.

Optionally, the first input information may also include at least one following:

Temporal information, spatial information, image collecting device identification information.

It should be noted that each image collecting device has the mark of a unique characterization described image acquisition device Know.

Here, the spatial information includes at least geographical location information.

Here, described image acquisition device has image collecting function, for example, described image acquisition device can be camera shooting Machine or capture machine.

Illustratively, the first input information can be inputted by civil servant such as police in terminal side, the terminal energy Enough to connect with system database, the system database is stored with the polymerization file data established based on clustering.

Here, the image of the target object can be is collected by image acquisition device such as video camera or camera etc. , it can also be and scanned by scanner, can also be and received by communicator.The embodiment of the present application is to target The acquisition modes of the image of object are not construed as limiting.

Step 102 captures the image collecting device to the target object in mesh based on the first input acquisition of information The N seconds candid photograph images in time point front and back are marked, the object time point is that described image acquisition device is captured to the target object Time point.

Wherein, the N is positive number.

It is described that the figure to the target object is captured based on the first input acquisition of information in an optional embodiment As acquisition device before and after object time point N seconds candid photograph images, comprising:

One or more image collecting devices are determined based on the first input information；

Obtain one or more of image collecting device acquired images or video；

The target image containing the target object is determined from described image or video；

On the basis of the target image, finds out from described image or video and adopted with the target image in same image N seconds candid photograph images before and after acquisition means lower object time point.

Specifically, one or more image collecting devices are determined according to spatial information.

For example, when spatial information characterizes the city A B cell, all video cameras in B cell are determined as to be verified Image collecting device.

For example, 10 video cameras are shared in B cell, target object X has been arrived in the candid photograph of video camera 1,3,9, then, it takes the photograph Camera 1 had photographed the image 1 containing target object X, and on the basis of the image 1, which was being photographed image 1 front and back N seconds in Acquired image is regarded as to be denoted as containing the candid photograph image of the same pedestrian of target object X and capturing library 1.Similarly, it takes the photograph Camera 3 had photographed the image 3 of target object X, and on the basis of the image 3, which is adopted photographing image 3 in front and back N seconds The image of collection is regarded as to be denoted as comprising the candid photograph image of the same pedestrian of target object X and capturing library 3.Similarly, video camera 9 The image 9 for having photographed target object X, on the basis of the image 9, the video camera 9 photographed it is collected in image 9 front and back N seconds Image is also regarded as to be denoted as comprising the candid photograph image of the same pedestrian of target object X and capturing library 9.So, may include The candid photograph image of the same pedestrian of target object X is formed by capturing library 1, capturing library 3 and capturing library 9, in step 103, is needed pair These three images captured in library are analyzed.

Step 103, the same pedestrian that the target object is determined from the candid photograph image.

In an optional embodiment, the same pedestrian that the target object is determined from the candid photograph image, packet It includes:

Determine the personnel in addition to target object occurred in capturing image；

The personnel in addition to target object are determined as to the same pedestrian of target object.

That is, searching the image collecting device captured to target object image N seconds M before and after object time point Width captures image, and the personnel in addition to target object occurred in this M width image are defined as to the same pedestrian of target object.

Step 104 analyzes the same pedestrian based on polymerization file data, obtains same pedestrian's recognition result, described The same person is corresponding with unique archives in polymerization file data.

In the embodiment of the present application, the polymerization file data is the system file number established based on clustering According to.The polymerization file data is stored in system database, and the system database is at least divided into first database and second Database；Wherein, the portrait figure that the first database is captured based on image collecting device at；The second database base It is formed in the image information of real name.

For convenience of understanding, the first database can be known as capturing portrait library, be arrived according to image collecting device candid photograph Portrait figure formed；Second database can be referred to as to static portrait library, according to citizen's population information of real-name authentication As identity card is formed.

It is described that the same pedestrian is analyzed based on polymerization file data in some optional embodiments, it obtains same Pedestrian's recognition result, comprising:

Wherein, with the relevant information of pedestrian, comprising:

For failing the same pedestrian of real name, relevant information is included at least in first database about each of the same pedestrian Capture image；

For the same pedestrian of real name, relevant information is including at least the image information in the second database, text information.

In this way, it is for statistical analysis to image is captured based on polymerization file data, it is capable of the same of quick obtaining target object The relevant information of pedestrian in this way, can help to search suspect partner, establish real name relational network, and then significantly facilitates investigation work.

In a specific example, terminal side obtains input information, and the input information includes suspect Q, and the period (can Be accurate to the second grade), camera identification, front and back t second, terminal side be based on the input information, find may comprising suspect Q colleague All candid photograph images of people are polymerize to image is captured, will be belonged to same based on the system database connecting with the terminal The candid photograph image of archives is aggregated to together.Terminal exports the correlation of all same pedestrians of suspect Q when receiving output order Information；Wherein, with the relevant information of pedestrian, it is specifically divided into the same pedestrian of the same pedestrian of real name and non-real name, it is specifically, real The same pedestrian of name includes: the text informations such as picture and identification card number, name, address, nationality in library；The same pedestrian of non-real name It include: to capture small figure.Here, the small figure of candid photograph is for capturing image, is the parts of images captured in image.

It is described that the same pedestrian is analyzed based on polymerization file data in some optional embodiments, it obtains same Pedestrian's recognition result, further includes:

Still by taking above-mentioned specific example as an example, terminal exports suspect Q when receiving the output order about colleague's number All same pedestrians colleague's number, and sequence exports from high to low or from low to high according to colleague's number.

Fig. 2 is colleague's number query result schematic diagram provided by the embodiments of the present application, as shown in Fig. 2, in query result circle In face, left side show colleague's number of people picture to this with pedestrian relevant the last 30 days captures frequency curve figure, at most captures the time Section column diagram captures the video camera position with pedestrian, and right side shows same pedestrian in colleague's number of different zones. In this way, the information such as colleague's number about same pedestrian are explicitly very clear, it can help to search suspect partner, establish and closed with pedestrian It is network, significantly facilitates investigation work.

It should be noted that it is appreciated that the information such as contents displayed on interface and layout, according to user demand or can design need It asks and is set or adjusted.

The first same pedestrian is determined from the colleague human sequence；

Here, described first is any one people in all same pedestrians with pedestrian.

In this way, on the basis of obtaining colleague's number target object can be inquired and individually with the note of going together in detail of pedestrian Record.

In a specific example, colleague number and same pedestrian of the terminal side in all same pedestrians for determining suspect Q Relevant information in the case where, receive input information, the input information include go together people G (be in all same pedestrians with pedestrian G One), terminal searches suspect Q and all records of going together with pedestrian G.Terminal when receiving output order, output Q with The relevant information that G goes together every time: the small figure of candid photograph, big figure, candid photograph time, video camera information including Q and G, and support to press and capture The mode of time sequencing and inverted order is ranked up display to result.Here, the small figure of candid photograph is for capturing image , it is the parts of images captured in image；Capturing big figure is for capturing small figure, is to capture image entirety.

That is, terminal supports following data query modes: target object archives ID+ mono- with pedestrian archives ID+ when Between range+video camera ID, Page sorting list query.

Fig. 3 be target object provided by the embodiments of the present application with individually with the record queries result schematic diagram of going together of pedestrian, As shown in figure 3, on the basis of Fig. 2 result schematic diagram, left side displaying target object with the candid photograph figure of pedestrian, capture the target Object and video camera affiliated area and video camera information with pedestrian, right side is shown goes together about the target object and same pedestrian Video.In this way, can help to search suspect partner, build about individually information is recorded explicitly very clearly with the colleague of pedestrian Vertical colleague's relationship network, significantly facilitates investigation work.

K same pedestrians are determined based on the colleague human sequence；Wherein, the K is positive integer；

Here, the K can be understood as the forward K same pedestrians that sort in colleague human sequence with pedestrian.

In this way, can count the K colleagues with pedestrian on the basis of obtaining colleague's number and record.

In this way, the K candid photograph numbers with pedestrian can be counted on the basis of obtaining colleague's record.

In a specific example, colleague number and same pedestrian of the terminal side in all same pedestrians for determining suspect Q Relevant information in the case where, receive input information, the input information includes that (preceding K for taking colleague's number most are same by TOP K Pedestrian, K can be unlimited), terminal counts candid photograph number of the TOP K of suspect Q with pedestrian under each video camera.Terminal is connecing When receiving output order, the candid photograph number of the same pedestrian of suspect Q under each video camera is exported.

That is, terminal supports following data query modes: multiple same pedestrian's archives ID+ time range+multiple-cameras ID counts the candid photograph number of video camera.

Fig. 4 is that colleague provided by the embodiments of the present application point query result schematic diagram occurs, as shown in figure 4, in Fig. 2 result On the basis of schematic diagram, left side show colleague's number of people picture to this with pedestrian relevant nearest 30 days candid photograph frequency curves figure, most It is capture period column diagrams, capture the video camera affiliated area with pedestrian more, right side be shown in marked on map it is each Candid photograph number under video camera.In this way, the candid photograph number about the same pedestrian under each video camera is explicitly very clear, can help It helps and searches suspect partner, determines and search network, significantly facilitate investigation work.

Obtain the specified video stream by specifying image acquisition device；

In this way, colleague's record of the same pedestrian of the TOP K occurred in designated source can be filtered out.

In a specific example, colleague number and same pedestrian of the terminal side in all same pedestrians for determining suspect Q Relevant information in the case where, receive input information, which includes that (first K 's TOP K for taking colleague's number most goes together People, K can be unlimited) and video source, the appearance point of the TOP K of terminal statistics suspect Q under designated source with pedestrian.Eventually It holds when receiving output order, what output occurred in designated source, the same pedestrian of suspect Q and TOP K in pairs The relevant information of colleague: it including Q and with the small figure of candid photograph of pedestrian, big figure, candid photograph time, video camera information, and supports to press and capture The mode of time sequencing and inverted order, is ranked up result.

That is, terminal supports following data query modes: target object archives ID+ more with pedestrian archives ID+ when Between range+multiple-camera ID, Page sorting list query.

Fig. 5 is the analysis result schematic diagram provided by the embodiments of the present application about single video source, as shown in figure 5, in Fig. 2 On the basis of result schematic diagram, left side shows designated source, the corresponding video camera information in the designated source, target object With the head portrait of same pedestrian, go together the time, right side is shown in the position of the corresponding video camera in the designated source marked on map It sets.In this way, analyze with pedestrian to single designated source, it can help to search suspect partner, determine and search network, it is big generous Just work is checked.

It should be noted that it is appreciated that the information such as display content and layout on interface, according to user demand or can set Meter demand is set or is adjusted.

Technical solution provided by the embodiments of the present application determines the same pedestrian of target object by capturing image, can quickly know It Chu not same pedestrian；By carrying out polymerization analysis to same pedestrian based on the polymerization file data in system, same pedestrian can be quickly determined Relevant information, be conducive to the recognition accuracy for improving same pedestrian.

Herein described technical solution can be applied to intelligent video analysis, the fields such as safety monitoring.For example, can be used for entering the room The cases investigations such as point, community's control are taken care of yourself by theft, anti-terrorism monitoring, medical trouble accident, strike of being involved in drug traffic, state.For example, case occurs Afterwards, there are the portrait photo of a suspect F in the police on hand, analyze in same pedestrian and upload suspect's photo in skill tactics, when crime is arranged Between section, in spot periphery d, can find and go together the Y above personnel's archives with suspect F, and then find partner track, And then confirm partner position；After finding partner's photo, and repeatable above step, find more possible partner's photos.In this way, Clue is connected convenient for the police, improves efficiency of solving a case.

In above scheme, before step 101, optionally, the method also includes:

Step 100 (not shown in figure 1): polymerization file data is established based on clustering.

In some optional embodiments, polymerization file data is established based on clustering, comprising:

In this way, whole archive informations of the available people in system.

In some optional embodiments, clustering processing is carried out to the image data in first database, comprising:

Go out face image data from the image data extraction in the first database；

In this way, give in numerous portrait candid photograph figures carry out face cluster method, i.e., by the set of face be divided by Multiple classes of similar face composition, by clustering the set that class generated is one group of data object, these objects with it is same Object in class is similar to each other, different with the object of other classes.

Fig. 6 shows the algorithm principle schematic diagram of face cluster provided by the embodiments of the present application, as shown in fig. 6, face is poly- The algorithm principle of class mainly includes three steps:

Step 1: new input feature vector and bottom library class center carry out nearest neighbor search, determine if to belong to by FAISS index In existing bottom library, i.e., whether there is classification.

Here, the FAISS is the abbreviation of Facebook AI Similarity Search, and Chinese is open source phase Class libraries is searched for like property.

Step 2: the processing to there is class another characteristic: with existing categorical clusters, and updating bottom library class center.

Step 3: the processing to unclassified feature: cluster determines classification, new cluster centre is added in the class of bottom library In the heart.

Fig. 7 shows the implementation process schematic diagram of face cluster provided by the embodiments of the present application, as shown in fig. 7, first determining Library is captured, then determines a feature to capture every picture in library, characteristic distance close (similarity is high) is aggregated in one kind, The picture captured in library is classified based on polymerization result.

Fig. 8 shows the result schematic diagram of face cluster provided by the embodiments of the present application, as shown in figure 8, each in left figure One feature of graphical representation indicates the photo captured, and shape similar representation similarity is higher；Right figure is by cluster The figure of reason carries out automatic cluster according to similarity, and one kind indicates a people.

In some optional embodiments, polymerization is carried out to the image data in the second database and handles to obtain polymerization processing As a result, comprising:

That is, identification card number is identical to be polymerized to an archives in the second database.

In some optional embodiments, the clustering processing result is associated point with the processing result that polymerize Analysis, comprising:

In this way, the corresponding identity information of the highest image of similarity is assigned to this class for capturing library, so that this kind of candid photograph Portrait real name.

In above scheme, optionally, the method also includes:

In this way, file data in system is updated or is supplemented in time when newly-increased increment occurs in database.

Fig. 9 shows archives Establishing process schematic diagram provided by the embodiments of the present application, as shown in figure 9, the process is mainly divided It is most of for storage, classification, association, one grade of a people, non-real name archives five.For portrait library, batch portrait storage will be same The portrait of identification card number is polymerized to an archives；For capturing for library, batch captures image storage or incoming video stream, timing Triggering cluster, such as one hour or one day cluster are primary, and the time is configurable, cluster for the first time for full dose, later increment cluster, and existing Some Type of Collective, without similar class can auto-polymerization at a new class.For increasing portrait newly, it can be put in storage in batches or individual enters Library, whether have with newly-increased portrait identical identification card number, if so, newly-increased portrait is polymerize if inquiring in the existing archives in portrait library Archives under common identity card number；If identification card number not identical with newly-increased portrait, new shelves are established for newly-increased portrait Case.It is captured for newly-increased, can be put in storage in batches or individual is put in storage or incoming video stream, showing for library is captured in clocked flip cluster, inquiry Have in archives whether to have and capture identical class with newly-increased, if so, by newly-increased candid photograph be aggregated to it is mutually similar under archives；If not yet Have and capture identical class with newly-increased, establishes new archives for newly-increased candid photograph, library is hit at the class center and portrait library of new class.Capture library with Library is hit in portrait library, specifically, captures divide into several classes (people) after the cluster of library, each class has a class center, a corresponding class Central feature value, each class central feature value carry out full dose 1:n comparison with portrait library again, take similarity highest TOP1 and be greater than One portrait of preset threshold assigns the corresponding identity information of the portrait of this TOP1 to this class for capturing library, so that this kind of grab Clap portrait real name.

As it can be seen that regarding the portrait library (static library) with citizenship as pattern library, capture in conjunction with by capture machine to having The face snap figure of space time information is clustered, will be doubtful same in face identification system using similarity two-by-two as judgment criteria The information of people is associated, so that a people has unique general file.From archives, it can show that the attribute of potential suspect is special Sign, behavioural characteristic etc..

In this way, carrying out conditional filtering from all clustered in (including real name, non-real name) archives, find out at the appointed time The same people in the designated source of range captures certain personnel's archive information that number is more than a certain specific threshold.Obtain its archive information Afterwards, user can find quickly according to suspect's figure information in some period in some region and it is same at front and back t seconds Qualified same pedestrian is captured image and polymerize by capable people；It can also be on the basis for the colleague's number for obtaining same pedestrian On, suspect Q and individually going together record in detail with pedestrian G can be inquired, with judge colleague's record of certain suspicion personnel with Colleague's relationship net.

The problem of being difficult to realize efficient automatic clustering under the scene of mass data compared with the existing technology, the application energy Magnanimity suspect in video monitoring can also be captured image and public security both someone by enough candid photograph image automatic clusterings by magnanimity Member's database information carries out efficiently auto-associating.Herein described technical solution finds target by the specified requirements of input The candid photograph image of all same pedestrians of object, and further same pedestrian candid photograph image is polymerize and (belongs to the same archives Candid photograph is aggregated to together), based on the archives of target object can analyze with pedestrian, further clarify colleague's relationship network, Efficiently information is captured to all same pedestrians to utilize.

The embodiment of the present application also provides a kind of information processing units, and as shown in Figure 10, described device includes:

First obtains module 10, for obtaining the first input information；Wherein, the first input information, which includes at least, contains The image of target object；

Second obtains module 20, and the image for being captured based on the first input acquisition of information to the target object is adopted Acquisition means N seconds candid photograph images before and after object time point, the object time point are that described image acquisition device is captured to institute State the time point of target object；

Determining module 30, for determining the same pedestrian of the target object from the candid photograph image；

Processing module 40 is obtained same pedestrian and identifies knot for being analyzed based on polymerization file data the same pedestrian Fruit, the same person is corresponding with unique archives in the polymerization file data.

As an implementation, the processing module 40, is also used to:

Wherein, with the relevant information of pedestrian, comprising:

As an implementation, the processing module 40, is also used to:

The first same pedestrian is determined from the colleague human sequence；

As an implementation, the processing module 40, is also used to:

Obtain the specified video stream by specifying image acquisition device；

As an implementation, the processing module 40, is also used to:

In above scheme, optionally, described device further include:

Archives establish module 50, are used for:

As an implementation, the archives establish module 50, are also used to:

Go out face image data from the image data extraction in the first database；

As an implementation, the archives establish module 50, are also used to:

It will be appreciated by those skilled in the art that in some optional embodiments, in information processing unit shown in Figure 10 Everywhere in manage unit realization function can refer to the associated description of aforementioned information processing method and understand.

It will be appreciated by those skilled in the art that in some optional embodiments, it is each in information processing unit shown in Fig. 10 The function of processing unit can be realized and running on the program on processor, can also be realized by specific logic circuit.

In practical application, above-mentioned first, which obtains module 10, second, obtains module 20, determining module 30, processing module 40 The specific structure for establishing module 50 with archives may both correspond to processor.The specific structure of processor can be central processing Device (CPU, Central Processing Unit), microprocessor (MCU, Micro Controller Unit), digital signal Processor (DSP, Digital Signal Processing) or programmable logic device (PLC, Programmable Logic The set of electronic component with processing function or electronic component such as Controller).Wherein, the processor include can Code is executed, the executable code is stored in a storage medium, and the processor can pass through the communication interfaces such as bus and institute It states in storage medium and is connected, when executing the corresponding function of specific each unit, read from the storage medium and run institute State executable code.The storage medium is preferably non-moment storage medium for storing the part of the executable code.

The first acquisition module 10, second obtains module 20, determining module 30, processing module 40 and archives and establishes module 50 can integrate corresponding to same processor, or respectively correspond different processors；When integrating corresponding to same processor, institute It states processor and module 10, second acquisition module 20, determining module 30,40 and of processing module is obtained using time-division processing described first Archives establish the corresponding function of module 50.

Information processing unit provided by the embodiments of the present application, by being polymerize based on polymerization file data to image is captured The mode of analysis is conducive to the recognition accuracy for improving same pedestrian to determine same pedestrian and with the relevant information of pedestrian.

The embodiment of the present application also describes a kind of information processing unit, and described device includes: memory, processor and storage On a memory and the computer program that can run on a processor, the processor are realized aforementioned any when executing described program The information processing method that one technical solution provides.

In the embodiment of the present application, the processor is realized when executing described program:

As an implementation, it is realized when the processor executes described program:

Wherein, with the relevant information of pedestrian, comprising:

The first same pedestrian is determined from the colleague human sequence；

Obtain the specified video stream by specifying image acquisition device；

Go out face image data from the image data extraction in the first database；

Information processing unit provided by the embodiments of the present application, by being polymerize based on polymerization file data to image is captured The mode of analysis is conducive to the recognition accuracy for improving same pedestrian to determine same pedestrian and the relevant information in relation to same pedestrian.

The embodiment of the present application also describes a kind of computer storage medium, and calculating is stored in the computer storage medium Machine executable instruction, the computer executable instructions are for executing invoice recognition methods described in foregoing individual embodiments.? That is can be realized any one aforementioned technical solution after the computer executable instructions are executed by processor and provide Information processing method.

It will be appreciated by those skilled in the art that in the computer storage medium of the present embodiment each program function, can refer to The associated description of information processing method described in foregoing embodiments and understand.

Herein described technical solution, automatically the candid photograph image by same people in video monitoring and existing static personnel's number It is combined according to library, clue is connected convenient for the police, improves efficiency of solving a case.For example, finding it according to same pedestrian when detection gang crime He is suspect；By analyzing the same pedestrian of suspect, the social relationships of suspect are understood, and then check its identity and row Track.

It should also be understood that each alternative embodiment enumerated herein is only exemplary, it is used to help art technology Personnel more fully understand the technical solution of the embodiment of the present application, and are not construed as the restriction to the embodiment of the present application, this field Those of ordinary skill can carry out various changes and replacement on the basis of each alternative embodiment described herein, should also manage Solution is a part of the embodiment of the present application.

In addition, the difference of emphasizing each embodiment is focused on the description of technical solution herein, it is same or similar Place can be referred to mutually, for sake of simplicity, no longer repeating one by one.

In several embodiments provided herein, it should be understood that disclosed device and method can pass through it Its mode is realized.Apparatus embodiments described above are merely indicative, for example, the division of the unit, only A kind of logical function partition, there may be another division manner in actual implementation, such as: multiple units or components can combine, or It is desirably integrated into another system, or some features can be ignored or not executed.In addition, shown or discussed each composition portion Mutual coupling or direct-coupling or communication connection is divided to can be through some interfaces, the INDIRECT COUPLING of equipment or unit Or communication connection, it can be electrical, mechanical or other forms.

Above-mentioned unit as illustrated by the separation member, which can be or may not be, to be physically separated, aobvious as unit The component shown can be or may not be physical unit；Both it can be located in one place, and may be distributed over multiple network lists In member；Some or all of units can be selected to achieve the purpose of the solution of this embodiment according to the actual needs.

In addition, each functional unit in each embodiment of the application can be fully integrated in one processing unit, it can also To be each unit individually as a unit, can also be integrated in one unit with two or more units；It is above-mentioned Integrated unit both can take the form of hardware realization, can also realize in the form of hardware adds SFU software functional unit.

Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above method embodiment can pass through The relevant hardware of program instruction is completed, and program above-mentioned can store in computer-readable storage medium, which exists When execution, step including the steps of the foregoing method embodiments is executed；And storage medium above-mentioned includes: movable storage device, read-only deposits Reservoir (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or The various media that can store program code such as CD.

If alternatively, the above-mentioned integrated unit of the application is realized in the form of software function module and as independent product When selling or using, it also can store in a computer readable storage medium.Based on this understanding, the application is implemented Substantially the part that contributes to existing technology can be embodied in the form of software products the technical solution of example in other words, The computer software product is stored in a storage medium, including some instructions are used so that computer equipment (can be with Personal computer, server or network equipment etc.) execute each embodiment the method for the application all or part. And storage medium above-mentioned includes: that movable storage device, ROM, RAM, magnetic or disk etc. are various can store program code Medium.

The above, the only specific embodiment of the application, but the protection scope of the application is not limited thereto, it is any Those familiar with the art within the technical scope of the present application, can easily think of the change or the replacement, and should all contain Lid is within the scope of protection of this application.Therefore, the protection scope of the application should be based on the protection scope of the described claims.

Claims

1. a kind of information processing method, which is characterized in that the described method includes:

Image collecting device based on the first input acquisition of information candid photograph to target object N before and after object time point The candid photograph image of second, the object time point are time point of the described image acquisition device candid photograph to the target object；

The same pedestrian is analyzed based on polymerization file data, obtains same pedestrian's recognition result, the polymerization file data The middle same person is corresponding with unique archives.

2. the method according to claim 1, wherein described carry out the same pedestrian based on polymerization file data Analysis, obtains same pedestrian's recognition result, comprising:

The relevant information of all same pedestrians is determined based on the polymerization file data；

Wherein, with the relevant information of pedestrian, comprising:

For failing the same pedestrian of real name, relevant information is included at least in system in first database about each of the same pedestrian A candid photograph image；

3. the method according to claim 1, wherein described carry out the same pedestrian based on polymerization file data Analysis, obtains same pedestrian's recognition result, further includes:

4. according to the method described in claim 3, it is characterized in that, described carry out the same pedestrian based on polymerization file data Analysis, obtains same pedestrian's recognition result, further includes:

The first same pedestrian is determined from the colleague human sequence；

Wherein, it is described colleague record includes at least: the target object with described first with pedestrian candid photograph image, capture when Between, the identification information of image collecting device.

5. according to the method described in claim 3, it is characterized in that, described carry out the same pedestrian based on polymerization file data Analysis, obtains same pedestrian's recognition result, further includes:

Wherein, colleague's record includes at least: the target object and the K with pedestrian candid photograph image, capture the time, The identification information of image collecting device.

6. according to the method described in claim 5, it is characterized in that, the method also includes:

Obtain the specified video stream by specifying image acquisition device；

It is a with the same of pedestrian that the target object and the K under the designated stream are searched from all colleague's records Row record.

7. according to the method described in claim 5, it is characterized in that, described carry out the same pedestrian based on polymerization file data Analysis, obtains same pedestrian's recognition result, further includes:

Based on the target object and the K all records of going together with pedestrian, statistics institute under each image collecting device State the K candid photograph numbers with pedestrian.

8. a kind of information processing unit, which is characterized in that described device includes:

First obtains module, for obtaining the first input information；Wherein, the first input information, which includes at least, contains target pair The image of elephant；

Second obtains module, for capturing the image collecting device to the target object based on the first input acquisition of information The N seconds candid photograph images before and after object time point, the object time point are that described image acquisition device is captured to the target The time point of object；

Processing module analyzes the same pedestrian for being based on polymerization file data, obtains same pedestrian's recognition result, described The same person is corresponding with unique archives in polymerization file data.

9. a kind of other device of information processing, described device includes: memory, processor and storage on a memory and can handle The computer program run on device, which is characterized in that the processor realizes that claim 1 to 7 is any when executing described program Information processing method described in.

10. a kind of storage medium, the storage medium is stored with computer program, and the computer program is executed by processor When, so that the processor perform claim requires 1 to 7 described in any item information processing methods.